[josm-dev] JOSMng prototype [resend]

[Petr Nejedly]

Gabriel Ebner napsal(a):

>> modification status. It uses as little memory as feasible (A node with no
>> tags - or "created_by tag only" - needs just a single heap object).
> Impressive, you're even sharing the tag values.  Next thing you'll tell me is
> that you've specialized a way class for ways with only two nodes. :-)

No :-) There aren't that many of them anyway.

>> File->Open (blocks UI for a while currently)
> 
> Albeit only half as long as JOSM.
Maybe, but the point was: it shouldn't block.

> BTW, I'm getting lots of "Average shift before rehash is 1.0227355" messages.
> Did you roll your own hashtable?

Yes, kind of. It is more of a HashSet with X getEqual(X) and a possibility
of using some inner state directly as a key. No $Entry instances. What you see
were my statistics for tuning the mixing function and load factor.

> Would it even be possible to rewrite JOSM under a BSD license now?  I'd
> thought most of the likely contributors would be too contaminated by GPL'd
> code.

While "contamination" is not relevant (most programmers don't remember code,
only ideas, and there are no earthshaking ideas in JOSM either), I'm fine with GPLv2

> There are some parts that would very likely need more than mere duplication,
> i.e. a major rewrite (or even redesign):
>   - Merging
Sure, but the algorithm (semantics) should mostly stay.
>   - Selection
The first thing no my list.
>   - Actions
Mostly complete rewrite. No Commands either.
>   - Uploading
At least the OSM API implementation might stay.
>   - Validator plugin
...

> However, I don't really feel comfortable with cutting-and-pasting your code
> into trunk and be done with it.  If we're doing a rewrite, why do it only
Impossible anyway. Visualization needs updates from dataset. Dataset is encapsulated
and won't let you touch the fields directly.

> In any case, I've got some time to spare during the following two or three
> weeks and would love to make JOSM snappier than ever.

OK, let's settle on the source code transfer first. I don't even have SVN access
now. BTW: I did the initial development inside a HG repo.

-- 
Petr "Nenik" Nejedly, NetBeans/Sun Microsystems, http://www.netbeans.org
355/113 -- Not the famous irrational number PI, but an incredible simulation!